Knots in B-splines are critical points where the polynomial pieces of the spline change, effectively marking the transition between different polynomial segments. For example, in a quadratic spline, the point where the formula changes, such as x=2, is identified as a knot. B-splines are defined within specific intervals determined by these knots, and the additional knots at the beginning and end serve to create more equations for solving coefficients. The "not a knot" condition can be applied to ensure that the polynomials at the ends of the spline are continuous, providing necessary equations to balance the system. Understanding knots is essential for effectively constructing and analyzing B-splines in computational applications.